Making sense of EST sequences by CLOBBing them

Background: Expressed sequence tags (ESTs) are single pass reads from randomly selected cDNA clones. They provide a highly cost-effective method to access and identify expressed genes. However, they are often prone to sequencing errors and typically define incomplete transcripts. To increase the amount of information obtainable from ESTs and reduce sequencing errors, it is necessary to cluster ESTs into groups sharing significant sequence similarity. Results: As part of our ongoing EST programs investigating 'orphan' genomes, we have developed a clustering algorithm, CLOBB (Cluster on the basis of BLAST similarity) to identify and cluster ESTs. CLOBB may be used incrementally, preserving original cluster designations. It tracks cluster-specific events such as merging, identifies 'superclusters' of related clusters and avoids the expansion of chimeric clusters. Based on the Perl scripting language, CLOBB is highly portable relying only on a local installation of NCBI's freely available BLAST executable and can be usefully applied to >95% of the current EST datasets. Analysis of the Danio rerio EST dataset demonstrates that CLOBB compares favourably with two less portable systems, UniGene and TIGR Gene Indices. Conclusions: CLOBB provides a highly portable EST clustering solution and is freely downloaded from: [http://www.nematodes.org/CLOBB].